Transcript Introductory genetics for veterinary students

QTN modulating the transcription
rate of a chromosome domain
encompassing PLAG1 control bovine
stature.
Michel Georges
University of Liège
Belgium
Holstein-Friesian X Jersey
intercross
500 traits
Plan

Mapping the QTL




Genetic identification of the QTN



Intact ORFs support regulatory pQTN
pQTN affects expression of PLAG1-encompassing domain
Reporter assays and EMSA support causality of pQTN in PLAG1-CHCHD7
bidirectional promoter
Identifying the causative gene


HT sequencing identifies 13 candidate pQTN
Exploiting haplotype diversity to eliminate 5/13 candidate pQTN
Functional analysis of the QTN




Stature
A QTL affecting stature maps to BTA14
L+LD fine-mapping defines a 750 Kb CI
Naturally occurring null allele excludes CHCHD7
Conclusions
Stature
Stature
 Human:



Paradigmatic Quantitative Trait
h2 80%
Quasi-infinitesimal architecture
 Dog:


5 loci explain nearly all the difference of stature between
breeds.
Cattle:




Auroch: 2m=> domestic cattle: 1.1-1.5m
Economically important trait
h2 25-80%
Many reported “QTL”
Plan

Mapping the QTL




Genetic identification of the QTN



Intact ORFs support regulatory pQTN
pQTN affects expression of PLAG1-encompassing domain
Reporter assays and EMSA support causality of pQTN in PLAG1-CHCHD7
bidirectional promoter
Identifying the causative gene


HT sequencing identifies 13 candidate pQTN
Exploiting haplotype diversity to eliminate 5/13 candidate pQTN
Functional analysis of the QTN




Stature
A QTL affecting stature maps to BTA14
L+LD fine-mapping defines a 780 Kb CI
Naturally occurring null allele excludes CHCHD7
Conclusions
A QTL affecting stature maps to
BTA14: HF x J F2 population
A QTL affecting stature maps to
BTA14: line-cross model
294 microsatellites
A note on genome-wide
thresholds
 Significant:
 Expected to be exceeded on average 1/20 genome
scans
 Expected not to be exceeded in a proportion e^(1/20)=0.049 of genome scans
 Suggestive:
 Expected to be exceeded on average 1/1 genome
scans
 Expected not to be exceeded in a proportion e^(1)=0.63 of genome scans
 Determine these thresholds by genome-wide
permutation
A note on F2 vs line-cross
models
 Is it reasonable to assume that the Q
and q alleles will be alternatively “fixed”
in the alternate F0 lines? Hence, that the
QTL genotype of all F1 animals is the
same? Especially when this is not the
case for the SNP or microsatellite
markers.
 This has lead to many erroneous
conclusions of imprinted QTL in animal
genetics
A note on F2 vs line-cross
models
A QTL affecting stature maps to
BTA14: ½-sib model
Across-family analysis – 1 QTL
8->56 μsat.
A QTL affecting stature maps to
BTA14: ½-sib model
Within family analysis – effects
Within family analysis – significance
A note on bootstrapping
Plan

Mapping the QTL




Genetic identification of the QTN



Intact ORFs support regulatory pQTN
pQTN affects expression of PLAG1-encompassing domain
Reporter assays and EMSA support causality of pQTN in PLAG1-CHCHD7
bidirectional promoter
Identifying the causative gene


HT sequencing identifies 13 candidate pQTN
Exploiting haplotype diversity to eliminate 5/13 candidate pQTN
Functional analysis of the QTN




Stature
A QTL affecting stature maps to BTA14
L+LD fine-mapping defines a 780 Kb CI
Naturally occurring null allele excludes CHCHD7
Conclusions
Factors limiting mapping
resolution
 Marker density
 Cross-over density
 Current (linkage)
 Historical (LD; limits)
 QTL detectance
Within “blocks”, LD is a function if
mutational sequence NOT distance
LD doesn’t decay « simply »
with distance
D’<1
5
2
1
4
12 3 4 5 6
3
6
D’=1
The International HapMap Consortium. Nature 379: 1299 (2005)
Factors limiting mapping
resolution
 Marker density
 Cross-over density
 Current (linkage)
 Historical (LD; limits)
 QTL detectance
Combined L+LD mapping
accounting for stratification
Modeling polygenic and locusspecific effects as random effects
A feeling for the « individual animal
model »: MVN distribution
Bivariate normal distribution of the
animal effects of closely related ids.
Bivariate normal distribution of the
animal effects of unrelated ids.
Hidden Markov Models
Basic principles
Transition probabilities
1
2
Hidden
States
Emission
Probabilities
1
2
3
3
Hidden Markov Modeling of
haplotypes
L+LD fine-mapping defines
780 Kb interval: F2 population
• + 925 SNPs
• LD  non inbred F0
Multipoint analysis – 1 QTL/2 QTL
Single-point analysis – 1 QTL
10% of phenotypic variance
- Mixed model including “animal effect”
- Hidden Haplotype States
The grand-daughter design
L+LD fine-mapping defines
780 Kb interval: outbred pop.
Substitution effects of
hidden haplotype states
1% of phenotypic variance
Multipoint analysis – 1 QTL/2 QTL
No unique haplotype associated
with Q or q
“q”
3% of phenotypic variance
“Q”
Plan

Mapping the QTL




Genetic identification of the QTN



Intact ORFs support regulatory pQTN
pQTN affects expression of PLAG1-encompassing domain
Reporter assays and EMSA support causality of pQTN in PLAG1-CHCHD7
bidirectional promoter
Identifying the causative gene


HT sequencing identifies 13 candidate pQTN
Exploiting haplotype diversity to eliminate 5/13 candidate pQTN
Functional analysis of the QTN




Stature
A QTL affecting stature maps to BTA14
L+LD fine-mapping defines a 780 Kb CI
Naturally occurring null allele excludes CHCHD7
Conclusions
HT sequencing of 780 Kb
interval: =>13 candidate QTN
 M&M:
 “Progeny-tested” chromosomes
of six F1 sires
 103 long range PCR products
 Sire-specific multiplex identifiers
(MIDs)
 Roche FLX
MASA: Converting a polygenic trait in a series of monogenic entities
HT sequencing of 780 Kb
interval: =>13 candidate QTN
 Results:
 Average 20-fold coverage / sire
 9,572 variants  π: 1/300
 14 candidate QTN  segregation
pattern compatible with QTL
genotype.
HT sequencing of 780 Kb
interval: =>13 candidate QTN
HT sequencing of 780 Kb
interval: =>13 candidate QTN
HT sequencing of 780 Kb
interval: =>13 candidate QTN
Using Hidden Haplotype States
instead of sires to increase QTL
detectance
Exploiting “NGS”
Plan

Mapping the QTL




Genetic identification of the QTN



Intact ORFs support regulatory pQTN
pQTN affects expression of PLAG1-encompassing domain
Reporter assays and EMSA support causality of pQTN in PLAG1-CHCHD7
bidirectional promoter
Identifying the causative gene


HT sequencing identifies 13 candidate pQTN
Exploiting haplotype diversity to eliminate 5/13 candidate pQTN
Functional analysis of the QTN




Stature
A QTL affecting stature maps to BTA14
L+LD fine-mapping defines a 780 Kb CI
Naturally occurring null allele excludes CHCHD7
Conclusions
Across breed haplotype
diversity => 8 candidate QTN
Across breed haplotype
diversity => 8 candidate QTN
Across breed haplotype
diversity => 8 candidate QTN
Plan

Mapping the QTL




Genetic identification of the QTN



Intact ORFs support regulatory pQTN
pQTN affects expression of PLAG1-encompassing domain
Reporter assays and EMSA support causality of pQTN in PLAG1-CHCHD7
bidirectional promoter
Identifying the causative gene


HT sequencing identifies 13 candidate pQTN
Exploiting haplotype diversity to eliminate 5/13 candidate pQTN
Functional analysis of the QTN




Stature
A QTL affecting stature maps to BTA14
L+LD fine-mapping defines a 780 Kb CI
Naturally occurring null allele excludes CHCHD7
Conclusions
Intact ORFs support
regulatory pQTN
Plan

Mapping the QTL




Genetic identification of the QTN



Intact ORFs support regulatory pQTN
pQTN affects expression of PLAG1-encompassing domain
Reporter assays and EMSA support causality of pQTN in PLAG1-CHCHD7
bidirectional promoter
Identifying the causative gene


HT sequencing identifies 13 candidate pQTN
Exploiting haplotype diversity to eliminate 5/13 candidate pQTN
Functional analysis of the QTN




Stature
A QTL affecting stature maps to BTA14
L+LD fine-mapping defines a 780 Kb CI
Naturally occurring null allele excludes CHCHD7
Conclusions
Expression analysis: M&M
 79 fetuses
 QRT-PCR (SYBR and/or 3’exonucl.)
 x/8 internal controls selected with
geNorm
 ≤ 4 amplicons/gene
The pQTN affect expression of
all genes in conserved domain
The pQTN affect expression of
all genes in conserved domain
Average: 20.86 ≈ 1.8
Allelic imbalance at (pre-)mRNA
level => transcriptional effect
Conservation of synteny
suggests domain “regulon”
Plan

Mapping the QTL




Genetic identification of the QTN



Intact ORFs support regulatory pQTN
pQTN affects expression of PLAG1-encompassing domain
Reporter assays and EMSA support causality of pQTN in PLAG1-CHCHD7
bidirectional promoter
Identifying the causative gene


HT sequencing identifies 13 candidate pQTN
Exploiting haplotype diversity to eliminate 5/13 candidate pQTN
Functional analysis of the QTN




Stature
A QTL affecting stature maps to BTA14
L+LD fine-mapping defines a 780 Kb CI
Naturally occurring null allele excludes CHCHD7
Conclusions
3/8 candidate pQTN affect
Phastcons elements
Reporter assays and EMSA
support 2 promotor pQTN
**
Reporter assays and EMSA
support 2 promotor pQTN
Reporter assays and EMSA
support 2 promotor pQTN
Reporter assays and EMSA
support 2 promotor pQTN
Plan

Mapping the QTL




Genetic identification of the QTN



Intact ORFs support regulatory pQTN
pQTN affects expression of PLAG1-encompassing domain
Reporter assays and EMSA support causality of pQTN in PLAG1-CHCHD7
bidirectional promoter
Identifying the causative gene


HT sequencing identifies 13 candidate pQTN
Exploiting haplotype diversity to eliminate 5/13 candidate pQTN
Functional analysis of the QTN




Stature
A QTL affecting stature maps to BTA14
L+LD fine-mapping defines a 780 Kb CI
Naturally occurring null allele excludes CHCHD7
Conclusions
Pick your favorite gene …
X
X
Pick your favorite gene …
X
X
Formal test for gene causality:
Distribution of rare variant
Σ=5%
Σ=17%
Formal test for gene causality:
reciprocal hemizygosity
Steinmetz et al. 2002
Formal test for gene causality:
quantitative complementation
Naturally occurring null allele
excludes CHCHD7
CHCHD7 cis-eQTLwith distinct
segregation vector (vs “pQTL”)
Naturally occurring null allele
excludes CHCHD7
“eQTN” is a donor
splice site variant
Naturally occurring null allele
excludes CHCHD7
Naturally occurring null allele
excludes CHCHD7
 Splice site variant affects transcript levels in
multiple (all?) tissues
 pQTL and eQTL have different segregation
vector
 pQTL effect on stature is same for 4
segregating sires
 eQTN has no significant “residual” effect on
stature
 No failure to quantitatively complement
No failure to quantitatively
complement
Naturally occurring null allele
excludes CHCHD7
 Splice site variant affects transcript levels in multiple
(all?) tissues
 pQTL and eQTL have different segregation vector
 pQTL effect on stature is same for 4 segregating sires
 eQTN has nosignificant “residual” effect on stature
 No failure to quantitatively complement
=> CHCHD7 can not be sole causative
gene
Plan

Mapping the QTL




Genetic identification of the QTN



Intact ORFs support regulatory pQTN
pQTN affects expression of PLAG1-encompassing domain
Reporter assays and EMSA support causality of pQTN in PLAG1-CHCHD7
bidirectional promoter
Identifying the causative gene


HT sequencing identifies 13 candidate pQTN
Exploiting haplotype diversity to eliminate 5/13 candidate pQTN
Functional analysis of the QTN




Stature
A QTL affecting stature maps to BTA14
L+LD fine-mapping defines a 750 Kb CI
Naturally occurring null allele excludes CHCHD7
Conclusions
Conclusions
 QTN modulating the transcription rate of a chromosome
domain encompassing PLAG1 control bovine stature
 Domestic animal populations have unique features
facilitating the genetic dissection of complex traits
(line-crosses, harems, reduced effective population size,
haplotype diversity)
 Haplotype sharing may not always be effective for the
identification of old QTN
 The QCA can be applied in outbred populations using
naturally occurring null alleles